DI3: Linking Blender and Data Science

Today we are thinking about two questions - one tactical and one strategic. The tactical question is what steps are needed to join data science calculations with Blender. The strategic one is how to move through a series of Minimum Viable Products (MVP's) to reach Blender data science goodness. [0:00] Why AI and Blender together - what's the target application. [1:20] Approaches to overcoming integration obstacles [2:35] Jupyter notebooks and Blender [3:50] Amazon Web Services for high power calculations [5:30] Difference in requirements for training AI versus implementing AI [7:20] Concept of Minimum Viable Product and application to our Blender software development [7:40] Concierge model of development - create bespoke manual solution first [9:00] Step two - Convert bespoke solution to AI-supported solution [11:15] Optimizing by data science for the audience as a separate MVP [12:10] Starting the Doing Imperfection YouTube Channel to host training around Blender and data science. [13:45] Aspects of AI pose estimation and bringing generated pose descriptions into Blender [16:15] What is best practice for sharing Blender addon repositories? [19:15] Programming style of clearly separating configuration data from process code [20:45] Bye from Neil, thank you for listening. Please consider subscribing.